Investigations into Ultra-Low-Power Underwater Imaging

Imaging underwater environments is crucial to advancing our understanding of marine organisms, climate change, marine geology, aquaculture farming, and underwater archaeology. Despite significant advances in underwater imaging, scalable and longterm imaging of underwater environments is still an open problem. One of the main challenges in scalably imaging the ocean is that existing underwater cameras are too power-hungry for long-term observations. Recent work on ultra-low-power underwater imaging has shown that in-situ wireless underwater imaging is possible using fully submerged battery-free cameras and acoustic backscatter. Even though this is a promising advance, enabling truly useful ultra-low-power underwater imaging remains difficult due to many challenges and constraints including poor image quality (due to marine snow, hazing, and lighting conditions), limited energy, limited memory and computational power, and low bandwidth of the acoustic channel. This thesis investigates the various challenges that efficient and ultra-low-power underwater imaging faces and offers directions for solving them. In particular, we first survey the various challenges of ultra-low-power underwater imaging. Subsequently, we offer three solutions for addressing these challenges. First, we propose a simple denoising/desnowing method for ultra-low-power underwater imaging that shows ∼ 2𝑑𝐵 improvement in the quality of the images while reducing the memory consumption by ∼ 17x when compared to the state-of-the-art systems. Second, we perform ultra-low-power underwater edge inference that is ∼ 19x more memory efficient when compared to the baseline model with comparable accuracies. Then, we propose a solution for enabling ultra-low-power color imaging that is ∼ 10x less power-hungry than the state-of-the-art battery-free underwater imaging system. We conclude by offering a path to integrating these solutions into future end-to-end ultra-low-power underwater imaging systems.

Name thesis:degree_name

Master

Department dc:contributor.department

Program in Media Arts and Sciences (Massachusetts Institute of Technology)

Grantor dc:publisher

Massachusetts Institute of Technology

Year dc:date.issued

2023

Author dc:creator

• Naeem, Nazish

Advisor dc:contributor.advisor

• Adib, Fadel

Statement dc:rights

• In Copyright - Educational Use Permitted
• Copyright retained by author(s)

Licence dc:rights.uri

• https://rightsstatements.org/page/InC-EDU/1.0/

Handle dc:identifier.uri

https://hdl.handle.net/1721.1/152645

OAI identifier oai:identifier

oai:dspace.mit.edu:1721.1/152645